The proposed research is designed to investigate the stability or change of nonsampling survey errors in a panel study of older adults. Specific sources of error that will be investigated are: (1) nonresponse; (2) "Don't know" responses; and (3) validity of responses, established by methods of (a) comparison against external validation criteria and (b) construct validity (assessed by multimethod-multitrait procedures). Currently very little is known about the stability of such survey errors and their effect on the analysis of and the conclusions from longitudinal data. We propose to do a follow-up survey to a cross-sectional regional survey of 1016 older adults (60 years old and older) and 475 young and middle-aged adults (between 20 and 59 years of age) in which these survey errors were systematically asessed. The original survey focused on a range of topic areas that are of central importance in social gerontology, such as social contacts, religious commitment, work and retirement, housing and neighborhood, financial resources, special problems, and physical and mental health. Concepts in each of these areas were measured by several different methods. Several other forms of data collection were also used to assess the level of errors, in addition to the survey proper. They include: (1) observations by the interviewer during and after the interview; (2) observations and inquiries by the interviewer about reasons why some eligible respondents cannot be interviewed and about their characteristics; (3) short-term follow-up interviews by telephone with nonrespondents and a random subset of the respondents; (4) examination of administrative records from municipal and state agencies, and voter registration rolls; (5) maps; and (6) reports from spouses. The proposed follow-up survey will attempt to interview all respondents from the original survey plus all the persons who could not be interviewed in the original survey, by using the same questionnaire and most of the supplementary data collection procedures as in the original survey.